Patient Profile
K-Extractor™: Transform health records to structured knowledge
Customer Scenario
Need
Summarize Electronic Health Records (EHR) of a patient to support decision making for treatment or insurance quoting. Analyzing EHR in bulk to find cohorts or recognize outliers for fraud detection.
Usage of various expressions to describe the same finding, for example, both 'feeling worn out' and 'exhausted' refer to 'fatigue'. Smallest details matter: findings can have attributes like dosage, onset, time course, etc. Negative findings and findings related to family history need to be recognized correctly.
Challenge
Lymba's Solution
Solution
K-Extractor trained for medical domain recognizes important medical concepts and their attributes in EHR. The key findings are grouped together and represented as a structured profile.
Input
Electronic Health Records.
Structured representation of symptoms, signs, conditions, diseases, taken medication, procedures, etc.
Output
Key Features
Ontologies:
K-Extractor recognizes concepts from existing medical lexical resources including UMLS Metathesaurus, SNOMED-CT, & Medline (health signs, symptoms, diseases, medication, etc) as well as socio-demographic characteristics and relations (race, gender, age, nationality, insurance, financial status, education level) and semantic relations between these concepts.
Attributes Recognition:
Recognition and linking of various attributes: onset, severity, time course, symptom quality, alleviating or aggravating factors.
Similarity Recognition:
The system resolves duplicates expressed with different words, groups related findings together.
Check out K-Extractor for more details on the product and its features.
Or request a demo with us: